qwen-mcp
An all-in-one MCP server that enables AI agents with local system capabilities like filesystem navigation, command execution, desktop automation, and a continuity memory suite for persistent cross-session project context.
README
Qwen-MCP — All-in-One MCP Server & Continuity Memory
A comprehensive, token-optimized Model Context Protocol (MCP) server that equips AI agents (Claude, ChatGPT, Qwen Chat, Cursor, Antigravity, etc.) with local system capabilities and cross-session Continuity Memory.
Features
1. Continuity Memory Suite (Cross-AI Session Persistence)
Enables seamless handoffs across AI clients (Claude web, ChatGPT, Qwen Chat) when quotas run out mid-task:
memory_init: Initializes/loads project workspace under~/.continuity/projects/<project_id>/and returnsbriefing.md.memory_read: Readsbriefing(default), specificsection(architecture,todo,file-map,decisions), orfulldump.memory_update: Updates memory sections incrementally (replaceorappend).memory_log: Appends timestamped events to append-onlylog.jsonl.memory_compact: Deterministic server-side compaction of historical logs into structuredbriefing.md.memory_list: Lists all active projects and token estimates.memory_status: Token breakdown and client tracking.
2. Filesystem & Navigation
read_file: Read file contents with optional line range.write_file: Write/overwrite files with automatic parent directory creation.append_file: Append content to files.delete_file: Delete files safely.copy_file: Copy files with overwrite protection.list_directory: Tree-view directory listing with file sizes and directory counts.create_directory: Recursive directory creation.delete_directory: Safe directory deletion (recursive support).rename: Move or rename files/directories.get_file_info: Metadata (size, permissions, timestamps).file_exists: Check path existence and type.search_files: Search text patterns across files.find_files: Find files/directories by glob or regex.
3. Command Execution & System Monitoring
execute_command: Execute PowerShell (Windows) or Bash (Unix) commands with streaming logs and configurable timeouts (default 10 min).list_processes: List running processes with PIDs and memory usage.kill_process: Terminate processes by PID.list_active_ports: Inspect listening TCP network ports.get_system_info: Inspect CPU, RAM, OS, uptime, and disk usage.get_env/set_env: Read and update session environment variables.
4. Desktop & GUI Automation
take_screenshot: Capture PNG screenshots of active monitors.mouse_move/mouse_click: Coordinate-based cursor movement and clicking.keyboard_type: Simulated keystroke typing into active windows.read_clipboard/write_clipboard: System clipboard text access.send_notification: Native desktop notification alerts.http_request: Fetch APIs (GET, POST, PUT, DELETE, PATCH) with formatted responses.
5. Bundled Skills (MCP Prompts & Resources)
- Bundles the
continuity-memoryskill directly inskills/continuity-memory/SKILL.md. - Exposes skills over the MCP protocol as MCP Prompts (
continuity-memory) and MCP Resources (skill://continuity-memory).
Quick Start
Installation
git clone https://github.com/YOUSEFISLAM-dev/qwen-mcp.git
cd qwen-mcp
npm install
npm run build
Running
node dist/index.js
Configuration
| Environment Variable | Default | Description |
|---|---|---|
ALLOWED_ROOT |
process.cwd() |
Semicolon/comma separated allowed directory roots for filesystem tools (e.g. C:/Users/youse;F:/workspace). |
CONTINUITY_ROOT |
~/.continuity |
Storage directory for Continuity Memory projects. |
LOG_LEVEL |
INFO |
DEBUG, INFO, WARN, or ERROR. |
Client Integration Examples
Qwen Chat / Claude Desktop (claude_desktop_config.json)
{
"mcpServers": {
"qwen-mcp": {
"command": "node",
"args": ["C:/path/to/qwen-mcp/dist/index.js"],
"env": {
"ALLOWED_ROOT": "C:/Users/youse",
"LOG_LEVEL": "INFO"
}
}
}
}
Testing
# Run test suite (69 tests across 12 suites)
npm test
# Build TypeScript
npm run build
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。